an improved version of the differential bias corrected boundary shift integral from baseline, which: a) has a new intensity normalisation step that normalises scan intensity by matching the intensities by tissue type (cerebrospinal fluid, grey matter, white mater) instead of being just based on the average whole brain intensity. this helps reduce artefactual change in the bsi that might be caused by change in contrasts between scans (e.g scanner changes) b) automatically calculates the bsi window based on k-means clustering of tissue intensities for each scan pair